IS IT DEMOGRAPHICS AND NOT REGIONAL CULTURE? In the post from 2008 that I linked to yesterday, I expressed some misgivings about the predictive value of my theory that regional cultural differences have an important effect on voters. On March 13, the same date that Michael Lind’s article advocating the regional culture theory was published on Salon, Sean Trende published an article on Real Clear Politics, which supports the rival theory that what appear to be geographic differences can be explained by ethnic, educational and religious factors. Trende’s article has an appealing map of the vote by county for Romney, Santorum and Gingrich. Trende has constructed a regression equation which estimates the vote in a county for Romney based on the county’s percentages of Latinos, college-educated voters, LDS (Mormon), evangelicals and African-Americans. Evaluating regression equations requires detailed analysis of alternative formulations, but this approach demonstrates the alternative approach to explaining regional voting.
Note, in any event, that both geographic explanations show the pitfalls of “momentum” theories. The order in which primaries are held has a lot to do with “momentum”.